Research
We study how post-transcriptional regulation sets cell state. We build BenchDrop-seq, a benchtop platform for single-cell long-read RNA-seq that needs no microfluidics instrument, and Bagpiper, the pipeline that turns its long reads into isoform-by-cell counts. We apply both to drug resistance in NRAS-mutant melanoma and to immune dysregulation in multiple sclerosis and EBV infection.
RNA beyond mRNA
A gene makes more than its mRNA. Its enhancer and promoter are transcribed, its introns carry small RNAs, its 3′ end has more than one stop, and the polymerase can run on past it. The schematic lays these RNAs out as coverage tracks along one locus.
BenchDrop-seq combines PIP-seq single-cell capture with Oxford Nanopore long-read sequencing. It needs no microfluidics instrument, so any lab with a thermocycler can measure isoforms in single cells. The protocols are open and scale from pilot experiments to large immune-profiling studies.
Long-read single-cell data raises problems that short-read tools were not built for: sequencing errors blur the cell barcode and UMI, and a read can fit several isoforms. Bagpiper handles each step. It finds and error-corrects the cell barcode and UMI in every read, aligns the cDNA to the transcriptome, collapses PCR duplicates that share a cell, UMI and set of compatible transcripts, resolves molecules that fit several isoforms by expectation maximization, and outputs an isoform-by-cell count matrix from the raw Nanopore data.
Isoform measurements are easiest to interpret next to gene-level atlases and surface-protein markers. We transfer cell identity from established short-read references onto long-read BenchDrop-seq data with a two-layer annotation, so BenchDrop-seq datasets can be compared directly with existing immune and tumor atlases.
Our R package Matisse computes percent spliced in (PSI) for each splicing event in each cell and stores it as an assay in the Seurat object, next to gene expression, so splicing can be plotted on the same UMAP and compared across the same clusters. Its documentation walks through long-read and short-read workflows.
The lab runs two biology programs with equal weight. In cancer, with the Villanueva Lab at Wistar, we use BenchDrop-seq on NRAS-mutant melanoma to find the drug-tolerant persister cells that survive pan-RAS inhibition and to track the RRAS isoform switch that accompanies resistance.
In autoimmunity, with the Lieberman Lab at Wistar, the same pipeline profiles immune cells from multiple sclerosis patients, captured with Parse Biosciences, and EBV-driven immune dysregulation in ex vivo PBMC models. The question is where splicing changes T cell and B cell fate. Both programs ask how post-transcriptional regulation sets cell state.
Foundations: earlier work, not active Wistar directions
Tools developed during Avi’s graduate work with Rob Patro (salmon, alevin, alevin-fry, RapMap, Terminus) and postdoctoral work with Rahul Satija (Seurat v5, Signac, scCUT&Tag-pro) are widely used in single-cell analysis. They are not active research directions at Wistar; the lab’s own program is the four pillars above. See publications for the full record.
Collaborations
We share experimental systems, clinical samples and analysis problems with these groups.
Publications
For a complete list of publications from the Srivastava Lab and its members’ prior work, see the publications page.
View all publications →